Superimposed channel training for MIMO relay systems

Based on the knowledge of instantaneous channel state information (CSI), the optimal source and relay pre-coding matrices have been developed recently for multiple-input multiple-output (MIMO) relay communication systems. However, in real communication systems, the instantaneous CSI is unknown and needs to be estimated at the destination node. In this paper, we propose a superimposed channel training method for MIMO relay communication systems. It is shown that to minimize the mean-squared error (MSE) of channel estimation, the optimal training sequence at each node matches the eigenvector matrix of the transmitter correlation matrix of the forward MIMO channel. Then we optimize the power allocation among different streams of the training sequence at the source node and the relay node. Simulation results show that the proposed algorithm leads to a smaller MSE of channel estimation compared with the conventional MIMO relay channel estimation algorithm.